摘要
介绍了文本分类的基本过程以及朴素贝叶斯和K近邻算法等基本分类方法,给出了基于覆盖的构造性神经网络分类算法,并将其与朴素贝叶斯和KNN作了实验比较。结果表明,该算法具有较好的分类性能,适合于处理大规模的文本分类任务,从而有效地克服了传统文本分类算法的不足。
Introduced the basic process of text categorization and several basic categorization methods such as naive Bayes and K nearest neighbor algorithm. Then it presented cover based constructive neural networks, and compared it with NB and KNN. Experimental resuits showed that this algorithm had better classification capability and it was fit for dealing with large- scale task of text categorization, accordingly it could overcome the disadvantage of traditional text categorization.
出处
《计算机技术与发展》
2007年第7期183-185,189,共4页
Computer Technology and Development
基金
973计划(国家重点基础研究)(2004CB318108)
安徽省自然科学基金项目(0504200208)
关键词
文本分类
神经网络
覆盖算法
基于覆盖的构造性神经网络
text categorization
neural networks
cover algorithm
cover based constructive neural networks
